]> git.ipfire.org Git - thirdparty/paperless-ngx.git/commitdiff
Working arround current TIKA Library Bugs - lint
authorSimon Siebert <simon@tuxedo.home.a17t.dev>
Thu, 6 Jul 2023 21:31:38 +0000 (23:31 +0200)
committerTrenton Holmes <797416+stumpylog@users.noreply.github.com>
Thu, 3 Aug 2023 16:55:10 +0000 (09:55 -0700)
src/paperless_tika/parsers.py

index d69c5947a163102a56a20a0ed3d9f1fabf8a7922..0ba59d3f6232cbd31fa6e1a7188c96e6cca028a8 100644 (file)
@@ -1,9 +1,6 @@
 import os
 from pathlib import Path
 
-import array
-
-import dateutil.parser
 import httpx
 from django.conf import settings
 from django.utils import timezone
@@ -56,7 +53,7 @@ class TikaDocumentParser(DocumentParser):
 
         try:
             with TikaClient(tika_url=settings.TIKA_ENDPOINT) as client:
-                with open(document_path, 'rb') as f:
+                with open(document_path, "rb") as f:
                     content = f.read()
                     parsed = client.tika.as_text.from_buffer(content, mime_type)
         except Exception as err:
@@ -71,7 +68,7 @@ class TikaDocumentParser(DocumentParser):
 
         tz = timezone.get_current_timezone()
 
-        self.date = timezone.make_aware(parsed.created,tz)
+        self.date = timezone.make_aware(parsed.created, tz)
         self.archive_path = self.convert_to_pdf(document_path, file_name)
 
     def convert_to_pdf(self, document_path, file_name):